Protagonist name choice board

A shortlist needs a decision, not another hundred names

You have reached the stage where three names could plausibly belong to the protagonist. Generating another batch may postpone the decision rather than improve it. One candidate works in an introduction, another fits a whispered request, and a third looks right beside the family name. The difficulty is no longer finding text. It is deciding which tradeoff serves the scene you are writing.

This board compares supplied candidates inside one shared scene frame. You record each direct preference and its reason, inspect the resulting pattern, and make an explicit author selection. The software does not infer a personality from a spelling, collect reader reactions or award artistic quality points. Its useful work is keeping your judgments visible enough to question.

The candidates are alternative drafts of one naming decision. They are not automatically three people in the cast. If you need names distributed across several roles, use the assignment workbench. If you have not established what the name must do, start with a naming brief. This page belongs between those activities: the requirements exist, a shortlist exists, and the author needs a defensible choice.

Original fictional scene: the clock that was allowed to pause

Original fictional scene: Oru Vale had repaired every mechanism in the little theatre clock except the wheel that moved its last door. The painted houses revolved, the wooden clouds parted, and the miniature shopkeeper raised a hand precisely when the bell should sound. Oru reached for the remaining wheel. The apprentice put a finger on the workbench, beside the part rather than on it.

“Oru Vale, can you leave one wheel unrepaired?”

Oru asked whether the apprentice had found another fault. No: the last door had always opened automatically, revealing the same painted garden. The apprentice wanted to turn that door by hand after each performance and choose a different ending. One evening the garden could remain. Another evening there might be a blank room, waiting for the audience to imagine what belonged there.

The repairer tried the handle. Without the automatic movement, the door could stay closed through a whole silence. That silence initially felt like an unfinished job. Then the apprentice opened the door onto a square of ordinary blue paper. A child sitting on the front bench called it a sea, and someone behind the child called it an empty sky. Oru set the wheel in a labelled drawer instead of fitting it back into the clock.

This is an invented theatre mechanism and a creative example, not a description of historical clockmaking or an instruction for repairing equipment. The naming decision concerns the apprentice’s request. In the writer’s draft notes, Taren Vale suggested a firmer address, Mio Vale offered a compact call, and Oru Vale supported the quiet pause. Those are this writer’s impressions of the alternatives, not meanings contained in the names.

If the repairer were called Taren Vale, the same request could still work. The writer might give the reply a longer hesitation so the character’s willingness to stop feels earned. Choosing Oru does not manufacture gentleness, and choosing Taren does not impose authority. A name can support an intended reading; action and dialogue still have to establish it. Apply the method by identifying the exact line where your naming choice matters, then asking what additional writing each alternative would need.

Give every candidate the same opportunity

Enter two to twelve candidate records with exactly id, name and note. The ID is a stable worksheet label, not a linguistic property. The default IDs lan, mio and oru identify the three drafts even though lan’s displayed name is Taren Vale. Do not edit an ID merely to make it look like the name; doing so changes the correspondence to stored judgments.

A name is a nonblank single line of at most 120 UTF-16 units. A note may be empty and is limited to 400. IDs start with an ASCII letter and use at most forty letters, digits, underscores or hyphens. Keep two alternative interpretations under different IDs if they deliberately share the same display name. The program preserves them separately instead of deduplicating your reasons by spelling.

The shared frame contains {name} exactly once and is limited to 1,200 UTF-16 units. The same literal frame is shown for every candidate. A replacement does not adjust pronouns, invent a nickname or rewrite the speaker’s response. Include the actual form being evaluated: a full name, a short address or another supplied candidate. If you need to compare a surname-only call, prepare that as the current naming decision rather than assuming the board has derived it.

Use the decision context to explain what the choice must serve. “The repairer learns to leave room for another person’s choice” is more actionable than “beautiful protagonist.” The context is limited to 300 units. It binds the comparison to an intention that can later change, so the board can distinguish an old judgment from a newly applicable one.

Three comparisons can tell three different stories

The default shortlist produces three unordered pairs. Open lan/mio, read both versions of the same request, and choose A, B, tie or defer. A and B follow the displayed ID order, not the order in which you happened to read the guide. Write a reason within 300 units. “Prefer Taren here because the apprentice’s request should challenge an established working habit” tells you more than “A sounds better.”

Next compare mio/oru independently. You might prefer Mio because a quick call suits the workshop exchange. Then compare lan/oru and prefer Oru because the final silence matters more than a firm initial address. The resulting declarations are lan over mio, mio over oru, and oru over lan. Every local reason can be sincere even though the preferences do not produce a single linear order.

A tie means that you currently prefer neither candidate over the other in this context. It is a completed comparison, not a missing answer. Defer records a reason for postponement and remains unresolved. An absent comparison is also unknown. The board never fills lan/oru from a lan/mio and mio/oru path, because your naming preferences need not be transitive.

The table retains each pair, its state and its recorded reason. Twelve candidates produce all sixty-six questions; the program does not silently stop at a convenient display limit. A shorter serious shortlist is usually easier to assess, but shrinking it should be an explicit author decision. Removing an inconvenient candidate can leave orphan records requiring attention; it is not an invisible way to obtain a finished board.

Read a cycle as a tradeoff you can investigate

The graph uses a directed edge only when a current comparison prefers one candidate. Ties and deferred questions are not preference edges. A cycle shows an actual route back to its starting candidate, accompanied by the IDs of the judgments used. The tool reports one concrete cycle witness for each mutually reachable cyclic group; it does not claim to list every possible circuit.

This is different from contradictory identity declarations. Preferences can reflect competing qualities. In the clock example, firmness, compact address and quiet ending have pulled the writer in different directions. The appropriate response may be to clarify which quality matters in the final scene, reconsider one comparison or accept the tradeoff in the final explanation. It is not automatically to discard whichever name has the fewest letters or the lowest imagined quality score.

“Preferred over every other candidate” has a narrow definition: a current direct win against each other supplied candidate. One missing comparison prevents that statement. A tie prevents a strict win against that opponent. A candidate with no recorded loss and no tie can still become such a winner if its remaining direct comparisons go in its favor. The possible-winner list states only that conditional possibility.

A completed board can have no universal winner. The win, loss, tie and pending counts summarize declarations, not an audience sample or a scientific attractiveness measure. Do not use a total of wins to claim that the tool has objectively identified the best protagonist name. Look back at the actual line and the reasons attached to its comparisons.

Try changing one assumption without changing the record

The what-if controls replace one comparison in a separate preview. Choose its pair, a hypothetical preference and a reason. In the three-name cycle, changing lan/oru to prefer lan makes Taren the direct winner while the actual board retains its cycle. The side-by-side table exposes that sensitivity without rewriting the writer’s real judgments.

This is useful when a choice seems to depend on one unresolved intention. You can ask what would happen if firmness mattered more than the final pause, then inspect the changed structure. The preview does not claim that your preference has actually changed. To commit it, return to the real pair and record a reasoned comparison there.

A trial retains its source candidates, scene and purpose. If those change, an old trial is marked stale or orphan and is not applied. That state does not by itself block an otherwise complete real decision memo. The real memo excludes the uncommitted trial; the private audit retains it so you can inspect what was tried. Neither route presents a hypothetical change as a reader response.

Make the final choice explicit

After reviewing every direct pair, choose a candidate in the selection controls and explain the tradeoff in at most 500 units. The program does not preselect the first record or automatically adopt a direct winner. You can choose Oru from the cyclic board because the final request matters more than the workshop’s brisk exchange, while acknowledging why the other candidates were attractive.

The complete memo requires current, nondeferred comparisons for all active pairs, no unresolved orphan or stale judgments, and a current explicit selection. A cycle does not prevent completion. A direct winner does not bypass missing comparisons among the remaining candidates. This deliberate scope makes the memo a record of the whole declared shortlist, rather than a certificate for an isolated favorite.

You may record a selection early as a draft, but subsequent comparison edits make it stale. The selection binds the complete candidate records, scene, purpose and actual judgments. This includes changed reasons, not only a changed winner. Revisit the final explanation after altering its basis. Restoring the exact old source context can restore validity, which means context matching rather than independent approval.

The complete download contains the original candidates, shared frame, actual reasons, preference structure, explicit choice and the chosen scene. It does not rename a manuscript, allocate names to other roles, clear rights or establish a name’s real cultural origin. Use the selected form in your draft, then continue to occurrence migration or production proofing if that change affects existing assets.

Recover from edits without hiding the old decision

Changing one candidate’s name or note stales the comparisons involving that record. Comparisons between unchanged candidates can remain current. Changing the common scene or purpose affects every pair. Removed endpoints remain visible as orphan judgments until explicitly removed; removing their records does not silently erase the reasoning that referred to them.

Old visible buttons also check the current form before recording. If you edit a source field without updating the board, the button refuses to capture its old context. It also refuses to overwrite a pair judgment changed behind that visible control. Update the board, inspect the current source and record the intended decision. A blank reason, malformed JSON or unsupported record field produces a clear failure rather than a plausible-looking inherited result.

Copying, comparison, favorites, project saves and exported audits contain private candidate notes and reasons. Same-label candidates can be saved independently because their records and context remain distinct. A project file restores an editable worksheet; a complete memo records a completed choice within its declared scope. Review either before sharing it with collaborators who should not see abandoned identities or plot intentions.

Is a preference cycle an error I must eliminate?

No. It can expose competing scene priorities. Inspect its actual reasons, revise a comparison if your intention changed, or explicitly acknowledge the tradeoff when selecting a candidate. The graph is not a factual identity constraint.

Why is a missing pair not filled from two other preferences?

The board does not assume transitive artistic preferences. A over B and B over C do not supply your direct judgment of A versus C. Record that comparison or leave it explicitly unresolved.

Does the what-if preview change my completed naming decision?

No. It occupies a separate graph, does not write a real comparison or selection, and is excluded from the complete memo. To commit a new preference, use the actual pair controls and revisit your final choice.

Can I choose a name with fewer direct wins?

Yes. Explain the scene-specific tradeoff in your author selection. Counts are not quality scores. Completion requires current direct comparisons and an explicit reasoned choice, not a universal winner.